基于遥感和PPS分层抽样的区域棉花面积估算
发布时间：2018-10-30 16:27
【摘要】：针对传统抽样调查工作面临着野外调查工作量大、资料时效性较低且难以满足人们对数据现势性的高要求等一系列缺点,以新疆棉花种植主棉区沙湾县、玛纳斯县、呼图壁县为研究区,结合遥感技术提出了一种基于PPS分层抽样的空间抽样设计方案,并将该方案用于研究区棉花种植面积的估算。结果显示,PPS抽样与分层抽样结合后极大地提高了PPS抽样反推总体的方法优势。分配样本时分别采用按每层辅助变量之和的期望的算术平方根与该层待抽样单位总数之积、每层辅助变量之和进行比例分配的2种分配方法,其对应的反推总体的估计量变异系数分别为0.008、0.009,相对误差分别为0.016、0.017,分层后的样本变异程度极低,为反推结果的高精度打下了基础。2种样本分配方式的棉花种植面积提取精度均高于94%。该方法不仅精度高,而且在实际操作中简单方便。
[Abstract]:In view of a series of shortcomings of the traditional sampling investigation work, such as the heavy workload of field investigation, the low timeliness of data and the difficulty in meeting the high demand for the present situation of data, etc., the main cotton growing area of Xinjiang, Shawan County and Manas County, is the main cotton growing area in Xinjiang. This paper presents a spatial sampling design scheme based on PPS stratified sampling in Hutubi County, which is based on remote sensing technology, and applies it to the estimation of cotton planting area in the study area. The results show that the combination of PPS sampling and stratified sampling greatly improves the advantage of PPS sampling. When the sample is allocated, two methods are used to distribute the sample according to the product of the arithmetic square root of the sum of the auxiliary variables in each layer and the total number of units to be sampled in that layer, and the sum of the auxiliary variables in each layer is allocated proportionally. The estimated coefficient of variation of the corresponding backstepping population is 0.008 / 0.009, and the relative error is 0.016 / 0.017, respectively. The variation of the sample after stratification is very low. The results laid a foundation for the high precision of the backstepping results, and the precision of cotton planting area extraction of the two methods of sample distribution was higher than that of 94%. This method not only has high precision, but also is simple and convenient in practical operation.
【作者单位】： 东华理工大学测绘工程学院;中国科学院遥感与数字地球研究所遥感科学国家重点实验室;国家统计局农村社会经济调查司;
【基金】：国家统计局新疆棉花种植面积遥感调查项目 国家自然基金项目(41371358) 国家“863”计划项目(2014AA06A511) 国家科技重大专项(14CNIC-032079-32-02)
【分类号】：S562;S127
，
本文编号：2300529
[Abstract]:In view of a series of shortcomings of the traditional sampling investigation work, such as the heavy workload of field investigation, the low timeliness of data and the difficulty in meeting the high demand for the present situation of data, etc., the main cotton growing area of Xinjiang, Shawan County and Manas County, is the main cotton growing area in Xinjiang. This paper presents a spatial sampling design scheme based on PPS stratified sampling in Hutubi County, which is based on remote sensing technology, and applies it to the estimation of cotton planting area in the study area. The results show that the combination of PPS sampling and stratified sampling greatly improves the advantage of PPS sampling. When the sample is allocated, two methods are used to distribute the sample according to the product of the arithmetic square root of the sum of the auxiliary variables in each layer and the total number of units to be sampled in that layer, and the sum of the auxiliary variables in each layer is allocated proportionally. The estimated coefficient of variation of the corresponding backstepping population is 0.008 / 0.009, and the relative error is 0.016 / 0.017, respectively. The variation of the sample after stratification is very low. The results laid a foundation for the high precision of the backstepping results, and the precision of cotton planting area extraction of the two methods of sample distribution was higher than that of 94%. This method not only has high precision, but also is simple and convenient in practical operation.
【作者单位】： 东华理工大学测绘工程学院;中国科学院遥感与数字地球研究所遥感科学国家重点实验室;国家统计局农村社会经济调查司;
【基金】：国家统计局新疆棉花种植面积遥感调查项目 国家自然基金项目(41371358) 国家“863”计划项目(2014AA06A511) 国家科技重大专项(14CNIC-032079-32-02)
【分类号】：S562;S127
，
本文编号：2300529
本文链接：https://www.wllwen.com/kejilunwen/nykj/2300529.html
